[Openmcl-devel] Is this a bug?

[Ron Garret]

On Feb 9, 2008, at 6:21 PM, Gary Byers wrote:

> The MOP says that DEFCLASS of an existing class uses
> REINITIALIZE-INSTANCE to change that class (in-place), with initargs  
> derived from the defclass form.

That's right.  But even the MOP leaves a little wiggle room about  
whether an unspecified option in defclass necessarily leads to an  
unspecified keyword argument in the subsequent call to reinitialize- 
instance.

 From the description of DEFCLASS:

"The default initargs class option, if it is present in the defclass  
form, becomes the value of the :direct-default-initargs keyword  
argument to ensure-class."

But it doesn't actually say what happens if :default-initargs is not  
specified.  There is room to interpret this as, "If default-initargs  
is not specified, then ensure-class is called with :direct-default- 
initargs specified as nil."

Also, if you really want to hew to an unspecified-in-defclass-means- 
unspecified (and hence sticky) -in-reinitialize-instance  
interpretation, then an unspecified metaclass would be sticky too, but  
it isn't:

? (defclass foo (ns:ns-object) () (:metaclass ns:+ns-object))
#<OBJC:OBJC-CLASS FOO (#xAADB810)>
? (defclass foo (ns:ns-object) ())
 > Error: Can't change metaclass of #<OBJC:OBJC-CLASS FOO (#xAADB810)>  
to #<OBJC:OBJC-METACLASS +FOO (#xAADB840)>.


and I would hope that anyone arguing that it should be would feel just  
a little queasy about it.

rg